Study On Calibration Based On Photogrammetry System And Surface Data Fusion Algorithm

With the development of modern industrial manufacturing,automobile,aircraft and other industry products mostly adopt irregular modelities,mould design,production and testing requires a lot of 3D digital modeling of the complicated surface and three-dimensional detection,these are the premise of quality control and product design.Therefore,industrial production urgently need rapid,accurate new 3D detection technology and reverse engineering design techniques.In this paper,the camera calibration technique based on the photogrammetric system is studied,compared and evaluated by serveral camera calibration methods based on the complex curved surface point cloud model.A method of camera calibration based on differential evolution particle swarm optimization is proposed.At the same time,we research data preprocessing method based on the characteristics of the local and global sampling for data reduction.This paper puts forward the improved ICP algorithm based on dynamic adjustment factor for three-dimensional point cloud data splicing method.In this paper,we focus on studying the surface data fusion algorithm,and propose the combination algorithm of immune genetic and ant colony to fit the surface and reconstruct the three-dimensional surface based on iterative and optimized fusion algorithm.Finally,the 3D model reconstruction algorithm of complete orthogonal V system is studied,and the precision of 3D model reconstruction is studied and compared with other algorithms.The main research content of the paper:1.According to problems of traditional camera calibration algorithm multi-parameters and calculation time-consuming,we study camera calibration technique.Differential evolution particle swarm hybrid optimization algorithm is adopted to improve the camera calibration.Each iteration is introduced into the differential evolution of mutation and crossover into the particle swarm optimization to maintain the diversity of population particles,and select the optimal particle in each iteration.Experimental results show that this algorithm can improve the convergence of the algorithm and prevent premature convergence of particles.Compared with GA,PSO and other algorithms,the mean square error of the calibration algorithm proposed in this chapter is only 0.169,and the calibration error result is smaller compared with other algorithms,and the algorithm performance is excellent.2.In view of the problem of the mass point cloud data of computer storage and slow processing speed,we propose the simplified algorithm based on local and global sampling characteristics,the non-uniform grid method is adopted to extract the local characteristics of point cloud data values.Then the global characteristics of the point cloud data is extracted by using the space voxel method.Finally,the formula is used to combine the two characteristic values.According to the experimental threshold ?(28)0.62,the optimal effect of the point cloud can be obtained.In view of problem of difficult mapping relationship of point to point and slow ICP convergence process,the algorithm is proposed to improve the ICP algorithm based on dynamic adjustment factor.The improved ICP algorithm can effectively reduce the number of iterations and time consuming,and improve the convergence accuracy of ICP,and maintain the stitching precision under the same initial conditions.3.For the ant system slower convergence speed,easily falling into local optima shortcomings,combined with the previous research foundation and characteristics of the surface fitting,we propose fusion algorithm based on immune genetic and ant colony,the immune genetic algorithm is introduced into the iteration of the ant colony algorithm,formed by all ants solution as initial population,the current global optimal solution as a vaccine for selection of initial population,crossover and adaptive vaccine operations by serveral iterations,comparing with iterative optimal solution and the current global optimal solution for optimal solutions.The experimental results show that the proposed fusion algorithm is better than the surface fitting effect of GA and IGA,error decreases 18% and 11% respectively.Time of the fitting method is less than GA and IGA,time reduce 21% and 8% respectively.It can be seen that this algorithmhas has a good convergence speed,the surface fitting time is less,and the fitting precision is higher.For node vector is difficult to determine as well as the data point parameterization problem,using combination of iterative and hybrid optimization algorithm(CIHO),this paper proposes an iterative methods point cloud based on parameter to determine the node vector,it is introduced into optimal weights and data points parameterization of projection optimization algorithm to solve the problem of surface reconstruction.Experimental results show that this method is fast,effective and robust.Compared with other algorithms,the algorithm is competitive in accuracy and time consuming.4.In view of the complete orthogonal functions system cannot introduce spctrum analysis into the problems geometry information processing field,we propose model reconstruction method based on complete orthogonal V system,with piecewise polynomial structure reconstruction of 3D model,the advatange of simple operation and applicable to any grid topology.Experimental results show that this method can achieve better reconstruction model and improve the reconstruction precision.The reconstruction model can be seen by three examples of reconstruction models,and the error of reconstructed models can be controlled in 0.25 mm,which meets the error requirement of the reconstructed surface.At the same time,the accuracy evaluation of the model reconstruction is studied,and smoothness and surface quality of the curve surface is analyzed.The experiments show that the number of control points has influence on the surface error,if the number of control points increase,the surface fitting precision improves.